
I've finally got a working ghc on Solaris. Its 6.4.1. I couldn't get the earlier, pre 6.4 versions going due to problems with my gcc. It seemed to be expecting a gcc shared library that my system didn't have, nor did any other gcc that I was able to compile. After loading a few prerequisites, and then reloading them with --enable-shared, its now going. I found a problem with a header file in the solaris distribution too - it expects stdint.h, which doesn't seem to be a part of Solaris anymore. I undef'd the HAVE_STDINT_H lines, and its working well now. Brian McQueen

Brian McQueen wrote:
I've finally got a working ghc on Solaris. Its 6.4.1. I couldn't get the earlier, pre 6.4 versions going due to problems with my gcc. It
Indeed, I cannot reliable use our ghc-6.2.x solaris installation any more after gcc-2.95.3 has been uninstalled.
seemed to be expecting a gcc shared library that my system didn't have, nor did any other gcc that I was able to compile. After loading a few prerequisites, and then reloading them with --enable-shared, its now going.
Which gcc version are you using? I use gcc-3.4.4. Do you mean you had to install the dynamic libraries libreadline.so.5, libncurses.so.5, and libgmp.so.3 ? (That was clearly stated on the distribution page: http://www.haskell.org/ghc/download_ghc_641.html#sparcsolaris) If you had to install something else, please tell us what it was.
I found a problem with a header file in the solaris distribution too - it expects stdint.h, which doesn't seem to be a part of Solaris anymore. I undef'd the HAVE_STDINT_H lines, and its working well now.
/usr/include/stdint.h is part of our solaris (2.8 and 2.10) installations (and therefore HAVE_STDINT_H was defined). I don't know where HAVE_STDINT_H matters. Christian

Yes, I had to install those specific version as stated on the page.
Also, I've got gcc-2.95 here and its a Sun Ultra 5.
Regarding stdint.h, ghci seems to run well, but when I tried to
compile some code that used the 6.4 libraries, the #includes were
broken due to there being no stdint.h on this platform.
Also there still is a problem of some kind, and it must have something
to do with the compiler. I've never seen this sort of thing out of a
compiler before. I'm compiling some small and fairly simple libraries
- the WASH package. For some reason, the compilation of
HTMLMonad98.hs is taking forever - and I mean forever. The file is
about 5000 simple lines long - nearly all data and instance
declarations. I killed it this morning after it was compiling that
single file for about 18 hours! Every other file compiled in minutes
or seconds! I don't imagine it could be due to bad code - bad code
shouldn't make a compiler go nuts. Any ideas about this?
Brian McQueen
On 9/29/05, Christian Maeder
Brian McQueen wrote:
I've finally got a working ghc on Solaris. Its 6.4.1. I couldn't get the earlier, pre 6.4 versions going due to problems with my gcc. It
Indeed, I cannot reliable use our ghc-6.2.x solaris installation any more after gcc-2.95.3 has been uninstalled.
seemed to be expecting a gcc shared library that my system didn't have, nor did any other gcc that I was able to compile. After loading a few prerequisites, and then reloading them with --enable-shared, its now going.
Which gcc version are you using? I use gcc-3.4.4. Do you mean you had to install the dynamic libraries libreadline.so.5, libncurses.so.5, and libgmp.so.3 ? (That was clearly stated on the distribution page: http://www.haskell.org/ghc/download_ghc_641.html#sparcsolaris)
If you had to install something else, please tell us what it was.
I found a problem with a header file in the solaris distribution too - it expects stdint.h, which doesn't seem to be a part of Solaris anymore. I undef'd the HAVE_STDINT_H lines, and its working well now.
/usr/include/stdint.h is part of our solaris (2.8 and 2.10) installations (and therefore HAVE_STDINT_H was defined). I don't know where HAVE_STDINT_H matters.
Christian

Brian McQueen
For some reason, the compilation of HTMLMonad98.hs is taking forever - and I mean forever.
See http://cvs.haskell.org/cgi-bin/cvsweb.cgi/fptools/ghc/compiler/utils/UniqFM.... particularly the comment for revision 1.42. It might be helpful for you to merge that patch into your copy of the compiler. Regards, Malcolm

Malcolm Wallace wrote:
Brian McQueen
writes: For some reason, the compilation of HTMLMonad98.hs is taking forever - and I mean forever.
See http://cvs.haskell.org/cgi-bin/cvsweb.cgi/fptools/ghc/compiler/utils/UniqFM....
particularly the comment for revision 1.42. It might be helpful for you to merge that patch into your copy of the compiler.
This patch is already part of (and included in) the ghc-6.4.1 (source and binary) distribution. (UniqFM.lhs rev 1.39.6.1) Still, "make" on WASH-2.3.1 takes a while due to the large files HTMLMonad98.hs and HTMLPrelude98.hs. Christian real 15m11.406s user 14m5.128s sys 0m22.841s maeder@leo -> uname -a SunOS leo 5.10 Generic_118822-18 sun4u sparc SUNW,Sun-Fire-280R
participants (3)
-
Brian McQueen
-
Christian Maeder
-
Malcolm Wallace